The trouble with GDP and emerging markets

Is Africa poor? Yes, compared to other regions of the world. But this apparently straightforward question is harder to answer than it seems. This shows why the technicalities involved in constructing GDP statistics matter a lot.

Specifically, take the question of whether or not Ghana is officially a poor country. Aid organizations use a threshold in terms of real GDP per capita set by the World Bank to designate whether a country is “low-income” or “middle-income,” and this in turn determines the kind of assistance it gets in aid and cheap loans. Until November 2010, Ghana was considered “low-income,” that is, a poor country. But between 5 and 6 November 2010, its GDP increased by 60% overnight, turning it officially into a “low-middle-income” country. The reality had not changed, but the GDP statistics had, because the country’s statistical agency had updated the weights used in calculating the price index, and consequently real GDP, for the first time since 1993.

Likewise, when Nigeria “rebased” its GDP calculations in 2014, it overtook South Africa to become the continent’s largest economy. A shift to take more account of booming industries such as mobile telecommunications and Nollywood movies, because of how much more significant they had become, increased the country’s GDP by 89% in one swoop. So Africa as a whole is probably not as poor as we’ve long thought. The trouble with using old weights is that the structure of the economy changes quite dramatically over time. In many African, Asian, and Latin American economies, the GDP calculations take no account of phenomena such as globalization, or the mobile phone revolution in the developing world.

Donors to the poor countries have been funding efforts to improve the way real GDP is calculated — there is an initiative known as PARIS21 (Partnership in Statistics for Development in the 21st Century) — but the planned improvements stretch out to 2020 and beyond. There are fundamental weaknesses with the collection of basic statistics such as what businesses there are, what they are selling, or what goods and services households spend their incomes on. The surveys needed to collect this information are carried out only infrequently. In fact, a recent study found that in the data set frequently used by economists to make international comparisons, 24 out of 45 countries had no price survey data at all.

Some countries are using weights that have not been changed since 1968, and only ten sub-Saharan African countries use weights less than a decade old. In each case, where old weights have been used for years, there will be large upward revisions in estimated real GDP when the weights are updated. This could profoundly change our impression of the character and weakness or strength of these economies; one estimate suggests that for twenty years, sub-Saharan African economies have been growing three times faster than suggested by the “official” data.

For this reason, the developed economies’ national accounts for the most part now use a “chain-weighted” price index in the calculation of real GDP, meaning that the weights used to combine the separate prices into one index change steadily year by year. Otherwise, as just described, the weights in any given base year would diverge further and further from the actual pattern of the economy.

The main catch with this method is that the “real terms” components of GDP no longer add up to the total: the equation C + I + G + (X – M) = GDP((consumer spending plus investment spending plus government spending plus exports less imports (the trade surplus or deficit)) no longer holds for the chain-weight inflation-adjusted figures.

Chain-weighting also tells a different big-picture story about the economy, just as rebasing does. For example, historic GDP statistics such as those developed by Angus Maddison for the Organization for Economic Cooperation and Development (OECD) have not been recalculated using chain weights. To do so would change the accepted picture of international growth patterns. Maddison noted, “Acceptance of the new measure for this period [pre-1950] would involve a major reinterpretation of American history.”

It would show U.S. productivity lower than the United Kingdom’s in 1914, for example, and much lower U.S. growth and level of GDP than the United Kingdom’s by 1929. This is certainly not the received wisdom among economic historians, which matters because their explanations of what drives growth—with great relevance for policies now—could turn out to be based on an inaccurate understanding of what the economy was “really” doing in the nineteenth and twentieth centuries.

Just as with developing countries now, changing the method of calculating a price index presents a different pattern of growth. The seemingly technical issue of how best to calculate a price index has some profound implications. In short, the choice of techniques completely alters even the broad outlines of the big picture on economic growth.

This is an extract from Diane Coyle’s book, GDP: A Brief but Affectionate History, published by Princeton University Press.